Clinical decision support for bipolar depression using large language models.

Management of depressive episodes in bipolar disorder remains challenging for clinicians despite the availability of treatment guidelines. In other contexts, large language models have yielded promising results for supporting clinical decisionmaking. We developed 50 sets of clinical vignettes reflecting bipolar depression and presented them to experts in bipolar disorder, who were asked to identify 5 optimal next-step pharmacotherapies and 5 poor or contraindicated choices. The same vignettes were then presented to a large language model (GPT4-turbo; gpt-4-1106-preview), with or without augmentation by prompting with recent bipolar treatment guidelines, and asked to identify the optimal next-step pharmacotherapy. Overlap between model output and gold standard was estimated. The augmented model prioritized the expert-designated optimal choice for 508/1000 vignettes (50.8%, 95% CI 47.7-53.9%; Cohen's kappa = 0.31, 95% CI 0.28-0.35). For 120 vignettes (12.0%), at least one model choice was among the poor or contraindicated treatments. Results were not meaningfully different when gender or race of the vignette was permuted to examine risk for bias. By comparison, an un-augmented model identified the optimal treatment for 234 (23.0%, 95% CI 20.8-26.0%; McNemar's p < 0.001 versus augmented model) of the vignettes. A sample of community clinicians scoring the same vignettes identified the optimal choice for 23.1% (95% CI 15.7-30.5%) of vignettes, on average; McNemar's p < 0.001 versus augmented model. Large language models prompted with evidence-based guidelines represent a promising, scalable strategy for clinical decision support. In addition to prospective studies of efficacy, strategies to avoid clinician overreliance on such models, and address the possibility of bias, will be needed.

Prioritizing Treatment Goals of People Diagnosed with Bipolar I Disorder in the US: Best-Worst Scaling Results.

Purpose: Bipolar I disorder (BP-I) is associated with significant disease burden, but evidence on treatment goals in people diagnosed with BP-I is scarce. This study sought to quantify treatment goals related to the pharmacological management of BP-I in adults in the US and to identify if subgroups of people with similar treatment goals exist. Patients and Methods: A best-worst scaling (BWS) of treatment goals was developed based on available literature and input from experts and patients and was distributed as part of a survey between August and September 2021. Survey participants were adults with a self-reported diagnosis of BP-I who were recruited via an online panel in the US. Participants were asked to prioritize the importance of 16 treatment goals using BWS. BWS scores were computed using multinomial logistic regression, with the scores across all goals summing to 100 for each participant. Subgroups of people with similar preferences were identified using latent class analysis. Results: The most important treatment goals for people diagnosed with BP-I (N=255) were "being less impulsive, angry, or irritable" (score: 9.73), or being "able to feel pleasure or happiness" (score: 9.54). Goals related to reducing the incidence of various potential adverse events of medication (scores: ≤4.51) or "reducing dependence on others" (score: 3.04) were less important. Two subgroups were identified. One subgroup (n=111) prioritized symptom-focused goals, considering "reducing frequency of mania, depression, and mixed episodes" and "being less impulsive, angry or irritable" the most important (scores: 12.46 and 11.85, respectively). The other subgroup (n=144) placed significantly more importance on social functioning-focused goals, including beginning or maintaining a relationship with a partner/significant other, and with family and/or friends (scores: 8.45 and 7.70, respectively). Conclusion: People diagnosed with BP-I prioritized emotional improvements. Subgroups of people with BP-I prioritized either symptom-focused or social functioning-focused treatment goals.

Treatment-resistant depression: definition, prevalence, detection, management, and investigational interventions.

Treatment-resistant depression (TRD) is common and associated with multiple serious public health implications. A consensus definition of TRD with demonstrated predictive utility in terms of clinical decision-making and health outcomes does not currently exist. Instead, a plethora of definitions have been proposed, which vary significantly in their conceptual framework. The absence of a consensus definition hampers precise estimates of the prevalence of TRD, and also belies efforts to identify risk factors, prevention opportunities, and effective interventions. In addition, it results in heterogeneity in clinical practice decision-making, adversely affecting quality of care. The US Food and Drug Administration (FDA) and the European Medicines Agency (EMA) have adopted the most used definition of TRD (i.e., inadequate response to a minimum of two antidepressants despite adequacy of the treatment trial and adherence to treatment). It is currently estimated that at least 30% of persons with depression meet this definition. A significant percentage of persons with TRD are actually pseudo-resistant (e.g., due to inadequacy of treatment trials or non-adherence to treatment). Although multiple sociodemographic, clinical, treatment and contextual factors are known to negatively moderate response in persons with depression, very few factors are regarded as predictive of non-response across multiple modalities of treatment. Intravenous ketamine and intranasal esketamine (co-administered with an antidepressant) are established as efficacious in the management of TRD. Some second-generation antipsychotics (e.g., aripiprazole, brexpiprazole, cariprazine, quetiapine XR) are proven effective as adjunctive treatments to antidepressants in partial responders, but only the olanzapine-fluoxetine combination has been studied in FDA-defined TRD. Repetitive transcranial magnetic stimulation (TMS) is established as effective and FDA-approved for individuals with TRD, with accelerated theta-burst TMS also recently showing efficacy. Electroconvulsive therapy is regarded as an effective acute and maintenance intervention in TRD, with preliminary evidence suggesting non-inferiority to acute intravenous ketamine. Evidence for extending antidepressant trial, medication switching and combining antidepressants is mixed. Manual-based psychotherapies are not established as efficacious on their own in TRD, but offer significant symptomatic relief when added to conventional antidepressants. Digital therapeutics are under study and represent a potential future clinical vista in this population.

Improving the pharmacotherapeutic treatment of agitation associated with bipolar disorder.

INTRODUCTION: Agitation is commonly encountered in people with bipolar disorder, particularly when experiencing a manic episode. The number of approved pharmacological agents to manage acute episodes of agitation in this population is limited. AREAS COVERED: A search was conducted using the US National Library of Medicine PubMed.gov resource for English-language papers of clinical trials and reviews/meta-analyses, using the text words 'bipolar disorder' AND 'agitation,' as well as any papers with both two text words in the title, without any date restrictions. EXPERT OPINION: Existing pharmacologic options approved by regulatory authorities for the treatment of acute episodes of agitation associated with bipolar disorder have similar degrees of efficacy but differ in their tolerability profiles and ease of use, giving clinicians an opportunity to individualize treatment. The goal is to treat mild-moderate agitation before it evolves into severe agitation, encouraging noninvasive pharmacologic treatment options. Inhaled loxapine and sublingual dexmedetomidine are newer options with rapid onset of action and may be preferable for patients willing to cooperate with treatment.

The Effect of Lurasidone on Anxiety Symptoms in Patients With Bipolar Depression: A Post Hoc Analysis.

Objective: To assess the effects of lurasidone on anxiety symptoms and sleep disruption, and their moderating and mediating roles on treatment response in bipolar depression.Methods: This post hoc analysis included pooled data from 2 previously published 6-week placebo-controlled trials of lurasidone for bipolar I depression conducted between April 2009 and February 2012. Hamilton Anxiety Rating Scale (HAM-A) "psychic anxiety" (items 1-6, 14) and "somatic anxiety" (items 7-13) subscores were calculated. Functional outcome was assessed by the Sheehan Disability Scale.Results: All subjects (n = 824) had at least 1 psychic anxiety and 729 (88.5%) had at least 1 somatic anxiety symptom at baseline. 594 subjects (72.1%) experienced baseline sleep disturbance. Lurasidone, as monotherapy (20-60 mg/d and 80-120 mg/d pooled dose groups vs placebo) and adjunctive therapy (20 to 120 mg/d flexibly dosed vs placebo) with lithium or valproate, significantly reduced HAM-A psychic anxiety (-4.82 vs -2.97, P < .001, monotherapy; -5.56 vs -4.26, P = .009, adjunctive therapy) and somatic anxiety (-1.89 vs -1.37, P = .048, monotherapy; -2.22 vs -1.47, P = .006, adjunctive therapy) subcomponents. Improvement in anxiety symptoms mediated reduction in depressive symptoms and functional impairment. Decrease in sleep at baseline predicted change in anxiety symptoms with lurasidone treatment at week 6.Conclusions: Lurasidone was superior to placebo in reducing psychic and somatic anxiety in the short-term treatment of bipolar depression. Improvement in depressive symptoms and reduction in functional impairment were associated with reduction in anxiety symptoms moderated by baseline sleep disturbance during lurasidone treatment.Trial Registration: ClinicalTrials.gov identifiers: NCT00868699 and NCT00868452.

Expanded Treatment Options and Addressing Unmet Needs in the Diagnosis and Treatment of Bipolar Disorder.

Bipolar disorder presents on a spectrum, with bipolar depression on one end and bipolar I on the other and a host of other presentations in between. In addition to its many permutations and the difficulty of differentiating between diagnoses, comorbidities, incorrect treatment, and low self-report contribute to delayed diagnoses and inappropriate or delayed treatment. Once a diagnosis is reached, the latest evidence of the safety and efficacy profiles of existing and emerging treatments adds to the complexity when developing treatment strategies for patients with bipolar disorder. As guidelines are updated and new treatments become available, developing individualized treatment regimens is key and collaboration between clinician and patient and family is critical in optimizing patient outcomes. New treatment options can reduce some of the side effect burdens associated with treating bipolar disorder, and clinicians should use measurement-based care to assess whether treatment changes are necessary, which requires engaging with the patient to monitor efficacy and manage side effects. It is important to ensure that the patient and family understand the information to foster informed decision making and create a better therapeutic alliance. Involving patients in designing their own treatment strategies according to their tolerability criteria can help combat the 90% nonadherence rate, and ultimately lead to better patient care.

Perspectives on the success rate of current antidepressant pharmacotherapy.

INTRODUCTION: There has been growing debate about the effectiveness of traditional antidepressants for the treatment of depression, and whether the clinical trials literature overstates the value of existing agents. Antidepressant efficacy is limited by suboptimal remission rates, lack of robust efficacy across diverse depressed subgroups, slow onset, and challenges managing tolerability. Clinicians can better navigate uncertainties in this area by recognizing patient-specific clinical and prognostic factors that influence the likelihood of antidepressant drug response. AREAS COVERED: The author summarizes pertinent literature regarding drug-placebo differences in antidepressant outcome as well as patient-specific factors that influence antidepressant drug responsivity across subtypes of depressive disorders. EXPERT OPINION: Standardized effect sizes for most monoaminergic antidepressants are relatively modest. At least one-third of treatment response derives from nonspecific (yet substantial) placebo effects, limiting the ability to compare antidepressant medication effects to that of 'no treatment.' Patients with high baseline depressive symptom severity are less likely to respond to placebo but may be more responsive to antidepressant pharmacotherapy than is the case in mild forms of depression. Patient satisfaction with antidepressant response must take into consideration not only efficacy for reducing symptoms but also drug tolerability/acceptability and tangible improvement in functional outcome and quality of life.

Expert consensus recommendations on the use of randomized clinical trials for drug approval in psychiatry- comparing trial designs.

The use of randomized clinical trials, in particular placebo-controlled trials, for drug approval, is the subject of long-standing debate in the scientific community and beyond. This study offers consensus recommendations from clinical and academic experts to guide the selection of clinical trial design in psychiatry. Forty-one highly cited clinical psychiatrists and/or researchers participated in a Delphi survey. Consensus statements were developed based on the findings of a published, peer-reviewed systematic review. Participants evaluated statements in two survey rounds, following the Delphi method. The expert panel achieved consensus on 7 of 21 recommendations regarding the use of randomized clinical trials. The endorsed recommendations were: (i) Results from placebo-controlled trials are the most reliable and (ii) are necessary despite the growing placebo-effect; (iii) it is ethical to enroll patients in placebo-arms when established treatment is available, if there is no evidence of increased health risk; (iv) There is a need to approve new drugs with the same efficacy as existing treatments, but with different side-effect profiles; (v) Non-inferiority trials incur an increased risk of approving ineffective medications; (vi) The risk of approving an ineffective drug justifies trial designs that incur higher costs, and (vii) superiority trials incur the risk of rejecting potentially efficacious treatments. The endorsed recommendations inform the choice of trial-design appropriate for approval of psychopharmacological drugs. The recommendations strongly support the use of randomized clinical trials in general, and the use of placebo-controlled trials in particular.

Development and validation of a new scale to measure chronic suicidal ideation: The Chronic Suicidal Ideation Inventory-5 (CSI-5).

OBJECTIVE: Little is known about characteristics that differentiate acute from chronic suicidal ideation. Studies have been hampered by the lack of a formal operational definition or measurement tool for chronic suicidal ideation (CSI). We sought to adapt a standardized measure of obsessive-compulsive thinking to rate chronic suicidal ideation in a mood disorder inpatient cohort. METHOD: We devised a novel self-report instrument, based on constructs related to obsessive-compulsive thinking and the Yale-Brown Obsessive-Compulsive Scale (Y-BOCS), as applied to preoccupations with suicidal thoughts and their persistence over 24 months in a previously described cohort of 97 inpatients with mood disorder. Exploratory factor analysis and principal components analysis were used to establish the factor structure of the instrument, and Cronbach's alpha and McDonald's omega were calculated to determine internal consistency of the scale. RESULTS: Significant Pearson correlations with individual scale items and total CSI-5 scores were evident in relation to the 2-, 6-, 12- and 24-month periods preceding the index hospitalization. A one-factor solution explained 75% of the variance in total CSI-5 scores. Cronbach's alpha was 0.914, and McDonalds's omega was 0.916. CSI-5 scores were not significantly associated with current depression severity scores or actual past suicide attempts. CONCLUSIONS: This 5-item adaptation of Y-BOCS-type questions provides an internally consistent and reliable assessment of chronic suicidal ideation in patients with mood disorder. The integrated assessment of time occupied by suicidal thoughts, activity interference, associated distress, efforts to resist suicidal thoughts, and degree of control over suicidal thinking provides a cohesive framework for understanding chronic suicidal ideation.

Effect of Sublingual Dexmedetomidine vs Placebo on Acute Agitation Associated With Bipolar Disorder: A Randomized Clinical Trial.

Importance: Acute agitation is common in patients with bipolar disorder and requires urgent management to relieve distress and to prevent escalation to aggressive behavior. Objective: To evaluate the effect of orally absorbed, sublingual dexmedetomidine, a selective α2A-adrenergic receptor agonist on symptoms of acute agitation in patients with bipolar disorder. Design, Setting, and Participants: Phase 3, randomized, double-blind, placebo-controlled trial conducted in 15 sites in the US with enrollment between February 24, 2020, and April 27, 2020, and final follow-up on May 21, 2020. A total of 380 adults with bipolar I or II disorder were randomized and 362 completed the study. Interventions: Participants were randomized to 3 groups: sublingual dexmedetomidine 180 µg (n = 127), sublingual dexmedetomidine 120 µg (n = 127), or placebo (n = 126). Main Outcomes and Measures: The primary efficacy end point was the mean change from baseline at 2 hours for the Positive and Negative Syndrome Scale-Excited Component (PEC) total score. The range of possible total scores is 5 (absence of agitation) to 35 (extremely severe). The secondary end point was the earliest time of a statistically significant change in PEC total score from baseline for the drug vs placebo. On the primary efficacy end point, to account for multiplicity associated with comparing 2 sublingual dexmedetomidine doses with placebo, the 2-sided significance level for each dose vs placebo was set at .025. Results: Of 380 patients randomized (mean age, 45.6 years; 54.8% women; and 56.1% Black individuals), 378 (99.5%) self-administered the study medication and completed the study. Baseline agitation was mild to moderate, with an overall mean PEC total score of 18.0. Two hours after taking the medication, the mean changes from baseline in PEC total score were -10.4 for sublingual dexmedetomidine 180 µg, -9.0 for sublingual dexmedetomidine 120 µg, and -4.9 for placebo. Least-square mean differences from placebo in the sublingual dexmedetomidine groups at 2 hours were -5.4 (97.5% CI, -6.6 to -4.2) for 180 µg and -4.1 (97.5% CI, -5.3 to -2.9) for 120 µg (both doses P < .001 vs placebo). Treatment effects began 20 minutes after taking the medication among patients in the sublingual dexmedetomidine groups (least-square mean difference for 180 µg, -1.1 [97.5% CI, -2.0 to -0.2]; P = .007; for 120 µg, -1.0 [97.5% CI, -1.9 to -0.1]; P = .009). Adverse events occurred in 35.7% of patients taking 180 µg of dexmedetomidine, 34.9% taking 120 µg, and 17.5% taking placebo. The most common adverse events (≥5%) in the respective 180 µg, 120 µg, and placebo groups were somnolence (21.4% and 20.6% vs 4.8%); dry mouth (4.8% and 7.1% vs 0.8%); hypotension (6.3% and 4.8% vs 0%); and dizziness (5.6% and 5.6% vs 0.8%). Conclusions and Relevance: Among patients with mild to moderate agitation associated with bipolar disorder, treatment with a sublingual film formulation of dexmedetomidine 120 µg or 180 µg, compared with placebo, resulted in significantly greater reduction in the agitation score at 2 hours. Further research is needed to understand the spectrum of patients for whom this treatment would be effective and feasible and to better understand the clinical importance of the observed effect size. Trial Registration: ClinicalTrials.gov Identifier: NCT04276883.